Back in the days when wireline telephony was pretty much the only pervasive network technology, we had perhaps sixteen thousand “edge” office locations in the US. Today, according to some information provided by the Wireless Infrastructure Association (WIA), the US has over 142 thousand towers, over 200 thousand “macrocell” sites, and almost half a million outdoor small cell nodes. Add that up and you get fifty times the number of access edge locations we had in the old wireline days, and that growth has some really important side effects.
One thing that’s really important is that fiber infrastructure deployment has decisively shifted toward the access network. It used to be that we used fiber primarily in aggregation missions, starting with those SONET rings and moving into the core of IP networks. Today, we deploy far more glass in the access network, and while fiber to the home isn’t universal and may never be, fiber close to the home is already a fact.
A second important thing is that mobile fiber backhaul is making it easier to create those close-to connections for FTTN. There’s already been pressure on operators to use fiber to home/business locations where demand density is locally high enough to pay back the pass costs. If you’re going to trench fiber to a node for mobile services, you could trench from there to a residential subdivision or a shopping mall at a lower cost, so more areas are open. You can also use mobile sites as FWA locations. As a result, wireline broadband for what would normally be classified as residential use is improving in reliability and capacity.
A third important thing is that we’re redefining “edge” in terms of access networking, at least, and raising the question of whether there will come a time when we could deploy service features and even edge computing outward beyond the current facility-based edge office sites. Probably, this won’t happen today or even next year, but we’re moving to a point where some features/functions might be hosted further out, providing we could make a business case for applications that depend on them.
One thing you probably see from my description of the side effects is that they’re “important” for what they might bring in turn, rather than what they are in detail. So what might they bring?
The most important result of these three edge points is that business-specific access is no longer a great opportunity for operators. Staying with the US market, the total number of multi-site businesses in the US is roughly 50,000 and these businesses have an average of about 32 satellite locations. Operators might have hoped to have over a million and a half prospective carrier Ethernet sites if everyone got on an IP VPN, but now we have to assume that at the minimum the primary market for carrier Ethernet would be the 50,000 main sites, and my data says that all but about ten thousand of these could easily be supported via the same fiber/PON connections used for residential broadband, just with different routing off the head end of the PON.
That, to me, means that operators have a dire choice to make almost immediately. Do they take steps now to condition residential-targeted fiber and FWA to support traditional MPLS VPN access, or do they accept that SD-WAN will eventually likely supplant MPLS VPNs completely and start deploying it instead? If they sit on their hands (a not-uncommon decision), they can expect to lose out to managed service providers or self-installed SD-WAN, or even to cloud-hosted SASE.
The second thing these three access-related points would do is to amplify what could be called uNode hosting opportunity over NFV hosting. Virtual functions in NFV were generally deployed in “cloud resource pools”, and when NFV ISG proof of concept trials looked at replacing CPE with VNFs, it rapidly reached the conclusion that you needed a generalized appliance to “host” something. Shift the mission from CPE to far-edge, and you get what I’m calling a “uNode”, a sort-of-white-box that’s designed to host small, agile, feature elements. I think that O-RAN already admits to the fact that you can’t expect the same sort of “function hosting” for the RAN as NFV had envisioned, and I think that we can expect to see the white box taking on new roles as it becomes the host de jure for far-edge feature elements.
The next of our access-related impacts is an increased interest in and value for “edge injection and interconnect”. We already know that “content injection” at the metro level (where it’s typically done today) results in traffic load congestion upstream toward the injection point. If we presume that many content experiences are real-time material, then we need to think about a better mechanism for distribution as that material increases. Some of this will likely be handled by my uNodes (creating another reason to have them at critical junction points in the access network), but some may also be handled by injecting things deeper in the access network. If we really see social-metaverse experiences develop, which eventually is likely, then that will likely require at least some interconnection of uNodes to adjacent metro centers to reduce latency.
Our final point is that access infrastructure changes will drive full convergence of mobile and wireline at the service level because of the increased harmony at the access infrastructure level. There have been operator-sponsored projects to address this for almost two decades (Project Icarus was one) but the unity of access infrastructure (and my uNodes!) will make it a lot easier. In fact, we can expect these points to create a full separation of access, transport, and services. That means continuity of calls, the Icarus goal, across wireline and wireless and even between devices you own, but it means a lot more.
These points reshape a lot of our industry, and it will also have to reshape regulatory policies and business models. Technically speaking, the Internet will become the universal dialtone for everything. All services, and I mean all services, will operate over the Internet and will thus become “OTT” services. That puts the infrastructure providers in a fatally compromised position, one that they can escape from only through some near-term effort to create a true service-over-the-Internet position for themselves, or by being rescued by regulators through settlements or subsidies.